Machine learning algorithms using the inflammatory prognostic index for contrast-induced nephropathy in NSTEMI...

Area of Science:

• Cardiology
• Nephrology
• Oncology

Background:

• The Inflammatory Prognostic Index (IPI) is linked to adverse outcomes in cancer patients.
• Contrast-induced nephropathy (CIN) is a significant complication in patients undergoing procedures with contrast media.
• Predictive markers for CIN in non-ST segment elevation myocardial infarction (NSTEMI) patients are crucial.

Purpose of the Study:

• To evaluate the predictive value of IPI for CIN development in NSTEMI patients.
• To develop and validate a nomogram incorporating IPI for CIN risk prediction.
• To assess the performance of machine learning algorithms in predicting CIN.

Main Methods:

• Retrospective analysis of 178 CIN (+) and 1511 CIN (-) patients.
• Development of a risk prediction nomogram including IPI.
• Application and evaluation of machine learning algorithms (Naive Bayes, k-nearest neighbors).

Main Results:

• Patients who developed CIN had significantly higher IPI levels.
• IPI was identified as an independent predictor of CIN.
• The IPI-based nomogram demonstrated good predictive ability and calibration.
• Naive Bayes and k-nearest neighbors were the most effective ML algorithms for CIN prediction.

Conclusions:

• IPI serves as a readily available and valuable biomarker for predicting CIN in NSTEMI patients.
• Machine learning algorithms, particularly Naive Bayes and k-nearest neighbors, enhance CIN prediction accuracy when incorporating IPI.
• The findings suggest potential for integrating IPI into clinical practice for early CIN risk stratification.